Hi, I am trying to use quantized model to do finetuning in order to improve accuracy,
but the original model is using floating point which means that there are problems with floating point and fixed point( from quantized model weights) , the error is showed below, please give me some suggestion on how to solve or let the finetuning is work:
below are error message Using pytorch 2.1.0 to do post-training static-quantization#####
Traceback (most recent call last):
File “Static_quantize5.py”, line 218, in
train(hyp, opt = opt,device = torch.device(‘cpu’), tb_writer = tb_writer)
File “/home/gene/YuRen/yolov7/train.py”, line 117, in train
model.load_state_dict(quantized_weights, strict=False)
File “/home/gene/miniconda3/envs/YuRen_copy1/lib/python3.8/site-packages/torch/nn/modules/module.py”, line 2152, in load_state_dict
raise RuntimeError(‘Error(s) in loading state_dict for {}:\n\t{}’.format(
RuntimeError: Error(s) in loading state_dict for Model:
While copying the parameter named “model.0.conv.weight”, whose dimensions in the model are torch.Size([32, 3, 3, 3]) and whose dimensions in the checkpoint are torch.Size([32, 3, 3, 3]), an exception occurred : (‘Copying from quantized Tensor to non-quantized Tensor is not allowed, please use dequantize to get a float Tensor from a quantized Tensor’,).
While copying the parameter named “model.1.conv.weight”, whose dimensions in the model are torch.Size([64, 32, 3, 3]) and whose dimensions in the checkpoint are torch.Size([64, 32, 3, 3]), an exception occurred : (‘Copying from quantized Tensor to non-quantized Tensor is not allowed, please use dequantize to get a float Tensor from a quantized Tensor’,).
While copying the parameter named “model.2.conv.weight”, whose dimensions in the model are torch.Size([32, 64, 1, 1]) and whose dimensions in the checkpoint are torch.Size([32, 64, 1, 1]), an exception occurred : (‘Copying from quantized Tensor to non-quantized Tensor is not allowed, please use dequantize to get a float Tensor from a quantized Tensor’,).
…there are many similar error about copying parameters
########
Please Help Me !! Thanks you very much!!!